15 Comments

Summary:

Android fragmentation appears to be diminishing, as 58.8 percent of devices that accessed the Android Market in the last two weeks are running either Android 2.1 or 2.2. But there’s another type of Android fragmentation that’s beyond Google’s control: the custom user interfaces from handset makers.

Nearly 60 percent of all Google Android devices that recently accessed the Android Market are running Android version 2.1 or 2.2, according to an online dashboard Google provides to developers. It took nearly nine months since Android 2.0 was introduced, but more than half of those frequenting the Android Market have the most recent and polished versions of Google’s mobile platform. Mashable notes that 18.9 percent of handsets still run the original Android 1.5 version, which is now four releases behind the most current version, known as Froyo. And as recently as March, Android devices were distributed fairly equally between versions 1.5, 1.6 and 2.x. Such fragmentation between different versions results in a varied user experience and applications that may not work across all Android devices.

While many consumers purchase a phone based on a hot new ad campaign and may not care about nor understand differences between the different Android versions, the fragmentation problem becomes evident when comparing different devices. My sister bought a Motorola Droid — then running Android 2.0 — just a few weeks before I bought my Nexus One in January. When she saw the interface tweaks on my handset, she wanted them. And she’s not the only one. I’ve read dozens of blog comments and forum posts of Android 1.5 or 1.6 handset owners wishing for an upgrade to gain the newer platform features and user interface optimizations. Eventually, the Droid did receive an upgrade to Android 2.1, but many handsets in the wild have no official upgrade path in their future.

From a developer’s standpoint, the shift towards a majority of phones running the most recent version of Android is a benefit. Each release of Google’s mobile operating system adds to the development API level by one, and programmers must specify the minimum API level required for any apps they code and release through the Android Market. API levels are forward compatible, but not backwards compatible. Effectively, platform fragmentation can lead to a developer coding software that only a small percentage of Android phones can use, which reduces potential audience and earnings.

The growth in devices running Android 2.1 is good for consumers and developers alike, but there’s another side to Android fragmentation that Google doesn’t control — custom interfaces, such as HTC Sense, Motorola’s Motoblur and Samsung’s TouchWiz. Each time Google updates Android, the handset-makers must spend development time and resources to re-integrate their customizations, which often leads to delays in platform upgrades — and additional fragmentation between Android versions. Without such control over the interface, consumers and developers still might find that their latest handset the most recent, nor is an application compatible with certain devices.

Related research on GigaOM Pro (sub req’d):

Google’s Mobile Strategy: Understanding the Nexus One

You’re subscribed! If you like, you can update your settings

  1. “Mashable notes that 18.9 percent of handsets still run the original Android 1.5 version”

    “Original” was 1.0 which shipped on ADP1s in December 2008

    1. Todd, good catch even though as far as I know, the 1.5 update is available for even the first devices. I’ll make a correction in the copy. Thx!

  2. Glenn Fleishman Wednesday, July 21, 2010

    The headline is overstating the case based on your source of data as a reliable indicator of devices by OS in the field.

    The Android Market is likely biased towards newer versions of the OS, given that there is a smaller set of software available (and unlikely to grow) for pre-2.0 devices.

    Growth in Android unit sales driven by 2.x devices also contributes to the skew, as well as the fact that people getting newer devices are more likely to check out the marketplace right away than people who have had devices for a while.

    Further, 1.x gear is still being introduced as “new,” including tablets. And there’s some large percentage of 1.5/1.6 systems that can’t be upgraded to 2.x. They lack the storage, minimum RAM, or other system resources.

    It will be interesting to see if fragmentation continues when 3.0 ships, whenever that’s planned. I suspect that 2.x->3.x will be a much smoother and broader upgrade process because 2.x devices are so much more capable than the early 1.x devices.

    1. I too considered the skew, but ultimately, it’s not that important.

      This is because in my view, this Marketplace analysis happens to produce the data most useful to developers who are interested in targeting an API development level that will offer the widest range of current potential users (and more importantly, purchasers). If it’s the case that many people using older devices are simply not accessing the market anymore (due to disinterest, lack of compelling apps, whatever), then to a developer they are an irrelevant group. If I were a developer, I’d be interested in the group of users that were most actively engaged in the marketplace — and if that happens to be people with new devices, then that’s what I’d want to target.

    2. Glenn, if you have a better source, I’m all ears. ;) I get your point, but at least we have some type of reliable, internal data to work with. And the actual percentage isn’t ultimately important — the trend that data provides is.

      As far as 1.x tablets, most (if not all) have no access to the Android Market, so they don’t impact this data. And as the headline is clear, we’re talking about handsets, not tablets.

      I hear you on the lack of updates for 1.5/1.6 devices that are still available. The whole situation irks me as a consumer and validates my choice of Nexus One this past January.

      1. Now you’re just being cute. This isn’t a matter of better or worse source; I’m sure there are Web analytics for aggregated major sites that tell a picture of platform usage by version.

        This is a matter of whether the Android Market provides the best gauge of platform fragmentation and trends. The trends are skewed in the way I describe, so you can’t rely on them to reflect accurately the larger trends.

        I suspect, but cannot prove, that the larger trends reflect the market version numbers, but I don’t buy that the market provides a basis on which to make a larger conclusion given the bias of who uses it.

        I mention tablets not to dispute the headline, but as another data point that 1.5/1.6 platforms are shipping.

        Google and its partners continue to pretend that there’s no fragmentation because Market differentiates clearly about what software is available, and thus people aren’t left behind. This disregards the many features that are added in each release that users want, and the requirement to upgrade phones to get new features, instead of the seeming brand/platform promise of upgrading the OS on the phone you own!

        Good move on a Nexus One. It’ll be a collector’s item.

      2. Guilty as charged — and you got me to fall off my chair laughing when I read your comment, so thanks for that! :)

        I hear you on the data and sources. The 3rd link in the post uses recent AdMob data (which of course, Google now owns) showing an equal distribution of versions between 1.5, 1.6 and 2.x. But that data comes from handsets access the web or apps that are tracked using AdMob code, i.e.: yet another incomplete picture.

        Believe me, I see a huge fragmentation issue, whether Google admits to it or not. I’ve probably written a half-dozen pieces on it this year. Part of the issue is systemic of the cellular market-at-large though: carriers make money when they sell handsets and get a service contract. They don’t make additional short-term revenue by building and pushing out a software update, so from a business-sense, it’s not in their best interest.

        Again, another reason that I’m sticking with my Nexus One, even though I like some of the recent new Android phones with larger displays.

  3. But this is phones which access the Android Market. It’s not necessarily the share of phones which run a version. Don’t be misled into thinking this necessarily represents what phones are out there being used. It might mean this; or it might not. If you had Android 1.x and there are no new apps being written for it, would you go to the Market? Now, if you have 2.x and lots of new apps keep appearing, you’re going to keep going back.

    The fact that the only phone running 2.2, the Nexus One, shows up as 3% should give you a clue that something’s wrong with this picture. Either the Nexus One did incredible business (in which case why stop selling it?), or its users were hitting the market because the fun stuff was there. Hmm, which is more likely?

    Hence my analysis: http://www.guardian.co.uk/technology/blog/2010/jul/21/android-version-market-share-google.

    Hell, even some download numbers from the Market would be nice.

    1. Charles,

      From your analysis: “True, Android sales have accelerated this year and 2.1 is getting more prevalent. But that comparatively big chunk of 2.2 accesses indicates, to me anyway, that this is a distorted picture of what handsets out there are truly running.”

      “That emphasizes the point, though, that there could be a big hinterland of Android 1.x users going “gah, nothing useful there” who visit rarely, while the (fewer) 2.x users are slamming the servers as fun stuff piles up.”

      These are all valid questions, but is there a use for the answers outside of journalist/academic circles? As stated in my reply above, to a developer, what use is attempting to factor in the unknown numbers of disenchanted, rarely visiting users? Why expend time coding compatibility for a group of users that visit the market so infrequently that they wouldn’t even appreciate your efforts?

      All a developer today needs to know is the audience of people who are most likely view/download/buy his app if he were launch it today. The chart provides that information and an indication of trends to come.

      1. @NeoteriX: “These are all valid questions, but is there a use for the answers outside of journalist/academic circles?”

        Accuracy, for one: the headline at the top of the page says “58.8% of Android handsets run a recent version”. Well, they might, but then again if you target that, rather than 1.5, you may be disappointed.

        “As stated in my reply above, to a developer, what use is attempting to factor in the unknown numbers of disenchanted, rarely visiting users? Why expend time coding compatibility for a group of users that visit the market so infrequently that they wouldn’t even appreciate your efforts?”

        Because if actually it’s not 41.2% but 75% (randomly chosen) of Android phones that are running 1.x, but don’t see any new apps because folk are developing for 2.x, then you’re leaving money on the table. You haven’t said what version you target. Do you aim for 1.x, or 2.x?

    2. Good analysis, Charles but I think NeoteriX raises a good point from a developer view.

      Bear in mind that some percentage of 1.5 and 1.6 devices have already churned, i.e.: folks leave them for another phone or better Android handset. And with 160,000+ activations per day, I’d take an educated guess that a growing percentage of those are or will be Android 2.1 devices.

      The 3% of devices running Android 2.2 maybe indicate some additional data skewing as you say, but I’m not sold on that point. Why? There is a reasonably large community creating Android 2.2 builds for rooted Android phones, meaning that not every Android 2.2 device is a Nexus One. Just last night I started looking into a Froyo build for an old G1, for example. :)

  4. Suck It Google Wednesday, July 21, 2010

    This so-called UI forking never happened with iOS because as Steve “The Smartest Guy in the Room” Jobs said: BEEN THERE BEFORE DONE THAT !!!

    Google may have a few smart s/w developers but Apple is in another level completely beyond the current capabilities of the Google Plex .

    Google is so short-sighted their Android platform cannot even support an industrial tablet of iPad caliber. They cannot support large screens and almost all of their Marketplace apps would break in the face of a superior set of hardware specs.

    iOS is much more advanced than the Linux/Java cobblestone that Google rushed to slop together without proper design. Some of Google’s internal dev projects use agile methodologies that have documented failure rates in excess of 60 percent which is among the highest in the history of computing. I could go on and on but just take my word that Google is massively over-rated.

    1. I personally like and use both platforms, but for all of your arguments, one fact still applies: there will be at least as many (if not more) Android handsets than iOS4 handsets activated this year. ;)

    2. I fully understand that my comment might get deleted or never pass admin approval but I have to say it.

      Really, Suck it Google? There’s nothing left for Google to suck cus the whole time I read your post all I heard was your mouth going slurp slurp slurp. I’ve never heard such a knowledgable sounding guy sound so Jobs sucking fanboi in my life.

  5. Haxed News » The State of the Android Ecosystem Thursday, July 22, 2010

    [...] Android OS versions 1.5 and 1.6 make up slightly less than 50% on Android devices, according to a GigaOm report, while versions 2.01 and 2.1 have steadily risen (most notably with the release of the Motorola [...]

Comments have been disabled for this post